Describe Image

The Describe Image node looks at an image with a vision model and writes a text description of it. You choose how much detail you want, or write your own question, and the node returns text for prompts, captions or alt text. The output is text, not an image.

When to use it

• Caption generated images for social posts or for accessibility.
• Turn an existing picture into a prompt, to make new images in the same spirit.
• Describe a reference image in labeled sections for later AI nodes.
• Write alt text automatically in a content workflow.
• Feed image descriptions into Generate Script to build a story around pictures.

Quick start

Inputs

The output, Text, is the description. It connects to any text input, such as the Prompt input of Generate Image, Combine Text or Generate Script.

Settings

When Custom Prompt (optional) is empty and you set pre or post text, that text becomes the whole instruction and replaces the built-in one.

When the node has more than one result, Result History in the settings panel lets you switch between them.

Start from a preset

Describe Image has factory presets for common jobs. Open the preset menu on the node and choose one:

Realism Audit returns a yes or no list of the signs that an image was generated, with a count.

Read Presets to save your own.

Credits

A run is one call to the vision model. The call is charged for the tokens it actually used, at the selected model's rate. The tokens are the image, the instructions and the description.

• Before the run, the Run button shows an estimate (≈ and a number) of a typical description on the selected model. Gemini 3.6 Flash costs less than the default, Claude Sonnet 5.5.
• A longer description costs more. Structured and Detailed write more text than Brief.
• A hold comes first. When the run starts, Nodaro holds enough credits for the longest answer the model may write. After the call, the run is charged what it used.
• Reasoning Effort has no price of its own. Advanced mode, and any reasoning effort on a Claude model, run the call on the maker's own service, at that service's rate.
• Failures. A failure that the model's maker billed is charged for that usage. A failure that was not billed is refunded.

The full rules are in How LLM calls are charged.

Tips

• Brief for prompts with context. Use Brief when the description feeds a node that has its own context, such as a script generator.
• Structured for later nodes. Use Structured when later nodes need the subject, the colors or the lighting separately.
• Detailed for captions. Use Detailed for flowing captions and alt text.
• Ask for what you need. A custom prompt such as "Describe the fashion items in this image, with colors and materials" gives focused results.
• Keep the default instructions if you need them. A custom prompt replaces the built-in instruction. To extend the default behavior instead, include those instructions in your own prompt.